This PDF is a selection from an out-of-print volume from the National Bureau of Economic Research Volume Title: Monetary Policy Rules Volume Author/Editor: John B. Taylor, editor Volume Publisher: University of Chicago Press Volume ISBN: 0-226-79124-6 Volume URL: http://www.nber.org/books/tayl99-1 Publication Date: January 1999 Chapter Title: Policy Rules for Inflation Targeting Chapter Author: Glenn Rudebusch, Lars E.O. Svensson Chapter URL: http://www.nber.org/chapters/c7417 Chapter pages in book: (p. 203 - 262) 5 Policy Rules for Inflation Targeting Glenn D. Rudebusch and Lars E. 0. Svensson 5.1 Introduction In this paper, we use a small empirical model of the US. economy to exam- ine the performance of policy rules that are consistent with a monetary policy regime of inflation targeting. In the real world, explicit inflation targeting is currently pursued in New Zealand, Canada, the United Kingdom, Sweden, Aus- tralia, and the Czech Republic. Inflation targeting in these countries is char- acterized by (1) a publicly announced numerical inflation target (either in the form of a target range, a point target, or a point target with a tolerance inter- val), (2)a framework for policy decisions that involves comparing an inflation forecast to the announced target, thus providing an “inflation-forecast targeting” regime for policy, where the forecast serves as an intermediate target (cf. Hal- dane 1998; King 1994; Svensson 1997a), and (3) a higher than average degree of transparency and accountability.’ Glenn D. Rudebusch is research officer at the Federal Reserve Bank of San Francisco. Lars E. 0. Svensson is professor of international economics at the Institute for International Economic Studies, Stockholm University, and a research associate of the National Bureau of Economic Re- search. The authors thank James Stock, Frederic Mishkin, and other conference participants for com- ments. The paper was also presented at the Federal Reserve Bank of San Francisco-Center for Economic Policy Research, Stanford University, conference on Central Bank Inflation Targeting, 6-7 March 1998. The authors thank Ben Bemanke, Carl Walsh, and other conference participants for comments. Part of the paper was written when Lars Svensson visited the Reserve Bank of New Zealand and Victoria University of Wellington. He thanks these institutions for their hospitality. The views expressed in the paper do not necessarily reflect the views of the Federal Reserve Bank of San Francisco, the Federal Reserve System, or the Reserve Bank of New Zealand. The authors thank Charlotta Groth and Heather Royer for research assistance, and Christina Liinnblad for sec- retarial and editorial assistance. 1. The rapidly growing literature on inflation targeting includes the conference volumes Leider- man and Svensson (1995), Haldane (1995), Federal Reserve Bank of Kansas City (1996), and Lowe (1997). See also the survey by Bernanke and Mishkin (1997). 203 204 Glenn D. Rudebusch and Lars E. 0. Svensson We model an inflation-targeting policy regime using loss functions over pol- icy goals. In our loss functions, inflation targeting always involves an attempt to minimize deviations of inflation from the explicit inflation target. In addi- tion, however, our inflation-targeting loss functions also allow concerns about real output (or more precisely about the variability of output because the natu- ral rate hypothesis is assumed). That is, we would argue there is no necessary connection between the specification of the loss function (other than that infla- tion variability must enter with a nonnegligible weight) and the specification of an inflation-targeting policy regime.* For support of this view, see, for ex- ample, the recent discussion by Fischer (1 996), King (1996), Taylor (1996), and Svensson (1996) in Federal Reserve Bank of Kansas City (1996).’ Thus we interpret inflation targeting as consistent with a conventional quadratic loss function, where in addition to the variability of inflation around the inflation target there is some weight on the variability of the output gap.4 In examining policy rules that are consistent with inflation targeting, we consider two broad classes of rules: instrument rules and targeting rules. An explicit instrument rule expresses the monetary policy instrument as an ex- plicit function of available information. We examine both optimal unrestricted instrument rules (a tradition that goes back at least to Taylor 1979; recent con- tributions include Blake and Westaway 1996) as well as optimal simple or re- stricted instrument rules, which involve only a few parameters or arguments (e.g., current inflation and output as in Taylor’s 1993 rule). However, no central bank, whether inflation targeting or not, follows an explicit instrument rule (unrestricted or simple). Every central bank uses more information than the simple rules are based on, and no central bank would voluntarily restrict itself to react mechanically in a predescribed way to new information. The role of unrestricted or simple explicit instrument rules is at best to provide a baseline and comparison to the policy actually followed. A targeting rule may be closer to the actual decision framework under infla- tion targeting. It is represented by the assignment of a loss function over devia- 2. One may argue, though, that the high degree of transparency and accountability serves to increase the commitment to minimizing the loss function, and to ensure that any concern about the real economy is consistent with the natural rate hypotheses and therefore reduces, or elimi- nates, any inflation bias. 3. As discussed in Svensson (forthcoming b), concerns about the stability of the real economy, model uncertainty, and interest rate smoothing all have similar effects under inflation targeting, namely, a more gradualist policy. Thus, if inflation is away from the inflation target, it is brought back to target more gradually (under “flexible” rather than “strict” inflation targeting, the inflation forecast hits the target at a horizon that is longer than the shortest possible). Svensson (1997b) argues that all inflation-targeting central banks in practice behave in this way, possibly with dif- fering weights on the different reasons for doing so. 4. Because inflation-targeting central banks, like other central hanks, also seem to smooth inter- est rates, our loss function also includes some weight on the variability of interest rate changes. 205 Policy Rules for Inflation Targeting tions of a goal variable from a target level, or deviations of an intermediate target variable from an intermediate target level (cf. Rogoff 1985; Walsh 1998; Svensson 1997a, forthcoming b). A targeting rule, combined with a particular model, is only an implicit instrument rule; typically, the equivalent of a first- order condition has to be solved in order to find the corresponding explicit instrument rule. (For an intermediate target variable that the central bank has complete control over, the first-order condition is trivial: equality between the intermediate target variable and the target level.) As an example, note that one interpretation of “inflation-forecast targeting” is that the policy instrument is adjusted such that a conditional inflation forecast (the intermediate target vari- able) hits the inflation target at an appropriate horizon. Combined with a par- ticular model, the instrument then becomes an implicit function of current in- formation; when the corresponding system of equations is solved for the instrument, the explicit instrument rule results. We shall examine several such targeting rules below. Our analysis proceeds as follows. Section 5.2 presents the empirical model we use, which is a simple two-equation model of U.S. output and inflation, somewhat similar to the theoretical model in Svensson (19974. The model captures some realistic dynamics (e.g., monetary policy actions affect output before inflation) in a very simple but tractable form. Section 5.3 first attempts to reduce the confusion caused by the literature’s use of two different meanings of “targeting” and then presents the different instrument and targeting rules we examine. Section 5.4 reports our results, with focus on output and inflation variability under a large set of various policy rules. We find that some simple instrument and targeting rules involving inflation forecasts do remarkably well in minimizing the loss function (relative to the optimal rule). Other policy rules, some of which are frequently used in the literature as representing infla- tion targeting, do less well. Finally, section 5.5 concludes. 5.2 An Empirical Model of U.S. Output and Inflation 5.2.1 Motivation Our choice of an empirical model of output and inflation is motivated by three considerations. First, we choose a simple linear model (as well as qua- dratic preferences below), so our analysis will be tractable and our results transparent. Our model consists of an aggregate supply equation (or “Phillips curve”) that relates inflation to an output gap and an aggregate demand equa- tion (or “IS curve”) that relates output to a short-term interest rate. Obviously, our model glosses over many important and contentious features of the mone- tary transmission mechanism. Still, we feel that the model has enough rich- ness-for example, in dynamics-to be of interest, especially when judged relative to some of the models used in previous theoretical discussions. 206 Glenn D. Rudebusch and Lars E. 0. Svensson Second, our model captures the spirit